Re: [COMMITTERS] pgsql: Add API functions to libpq to interrogate SSL related stuff. - Mailing list pgsql-hackers

From Heikki Linnakangas
Subject Re: [COMMITTERS] pgsql: Add API functions to libpq to interrogate SSL related stuff.
Date
Msg-id 54D1CBB5.6030300@vmware.com
Whole thread Raw
List pgsql-hackers
On 02/04/2015 02:23 AM, Tom Lane wrote:
> Heikki Linnakangas <heikki.linnakangas@iki.fi> writes:
>> Add API functions to libpq to interrogate SSL related stuff.
>
> This patch is one large brick shy of a load: it creates exported libpq
> functions but fails to ensure they always exist.  That's why jacana is
> unhappy; though TBH I'm astonished that any non-ssl-enabled builds
> are passing.  Apparently missing library functions are less of a hard
> error on Linux than they ought to be.

Yeah, that's surprising. I don't see any option in ld man page to make 
it warn either.

This also shows that we don't have any regression tests for this 
function. It's trivial, so I'm not worried about that, but in general it 
would be good to have a regression tests suite specifically for libpq. 
There are probably many other things that are not exercised by psql.

> I think probably the exported functions need to be defined in fe-exec.c
> or fe-connect.c, with bodies along the lines of
>
> #ifdef USE_OPENSSL
>     call OpenSSL-specific function
> #else
>     return NULL
> #endif
>
> (or whatever's appropriate when no SSL support).  We do want these
> functions to exist even in non-SSL-enabled builds.

Sure. There are dummy versions of all the other SSL-related functions, I 
just missed PQsslAttributes. The OpenSSL-versions are in 
fe-secure-openssl.c, and the dummy ones are in fe-secure.c, within a 
"#ifndef USE_SSL" block.

Fixed now.

- Heikki




pgsql-hackers by date:

Previous
From: Jim Nasby
Date:
Subject: Re: Fetch zero result rows when executing a query?
Next
From: Kyotaro HORIGUCHI
Date:
Subject: Re: pg_basebackup may fail to send feedbacks.